THE CHALLENGES OF IMPROVING REVENUE-RECOGNITION
STANDARD FOR MULTIPLE-ELEMENT FIRMS:
EVIDENCE FROM THE SOFTWARE INDUSTRY (SOP 97-2)
A Dissertation
by
ANUP SRIVASTAVA
Submitted to the Office of Graduate Studies of
Texas A&M University
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
May 2008
Major Subject: Accounting
THE CHALLENGES OF IMPROVING REVENUE-RECOGNITION
STANDARD FOR MULTIPLE-ELEMENT FIRMS:
EVIDENCE FROM THE SOFTWARE INDUSTRY (SOP 97-2)
A Dissertation
by
ANUP SRIVASTAVA
Submitted to the Office of Graduate Studies of
Texas A&M University
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
Approved by:
Chair of Committee,
Committee Members,
Head of Department,
Edward P. Swanson
Ekkehart Boehmer
Mary Lea McAnally
Senyo Tse
James J. Benjamin
May 2008
Major Subject: Accounting
iii
ABSTRACT
The Challenges of Improving Revenue-Recognition Standard for Multiple-Element
Firms: Evidence from the Software Industry (SOP 97-2). (May 2008)
Anup Srivastava, B.Tech, IIT Delhi; M.B.A., University of Delhi
Chair of Advisory Committee: Dr. Edward P. Swanson
I investigated whether implementing SOP 97-2, the revenue-recognition standard
for the software industry, reduces earnings informativeness. This standard is particularly
important for two reasons: First, its provisions coincide with provisions of SAB 101, the
current general revenue-recognition standard. Second, the software industry provides a
laboratory setting for examining multiple-element firms, whose revenue-recognition
challenges keep mounting as more and more firms bundle multiple products and
services. I found that implementing SOP 97-2 leads to additional revenue deferrals and a
decline in earnings informativeness. However, the market prices these deferrals as
revenues, as if these amounts had not been deferred. Moreover, the proforma earnings,
which I calculated by undoing the revenue deferrals, more strongly correspond with
market returns than do the reported earnings. My findings indicate that the accounting
numbers calculated using the pre-SOP 97-2 revenue-recognition rules more strongly
correspond with market returns than do those calculated using SOP 97-2. My findings
should interest FASB in its project on developing a new revenue-recognition standard.
iv
ACKNOWLEDGEMENTS
I thank my dissertation committee Ekkehart Boehmer, Mary Lea McAnally,
Edward. P. Swanson (Chair), and Senyo Y. Tse for their guidance and comments. I also
thank Anwer Ahmed, Michael Drake, Thomas Omer, James Myers, Linda Myers, Nate
Sharp, Elizabeth Tebeaux, Connie Weaver, and the workshop participants at Texas
A&M University for helpful comments. I also thank Deloitte & Touche Foundation for
doctoral fellowship.
v
TABLE OF CONTENTS
Page
ABSTRACT ..............................................................................................................
iii
ACKNOWLEDGEMENTS ......................................................................................
iv
TABLE OF CONTENTS ..........................................................................................
v
LIST OF TABLES ....................................................................................................
vii
SECTION
I
INTRODUCTION .......................................................................................
1
II
BACKGROUND INFORMATION ON SOP 97-2 .....................................
7
III HYPOTHESIS DEVELOPMENT ..............................................................
11
IV SAMPLE SELECTION, RESEARCH DESIGN AND RESULTS ............
14
Sample Selection.......................................................................................
Normal and SOP 97-2-Created Deferred Revenue ...................................
Descriptive Statistics.................................................................................
Univariate Tests on Deferred-Revenue Accounts ....................................
Testing Hypothesis H1..............................................................................
Testing Hypothesis H2..............................................................................
Testing Hypothesis H3..............................................................................
Supplementary Question 1: Naïve Reaction .............................................
Supplementary question 2: Aggressive Revenue-Recognition Before
SOP 97-2 Was Enacted .............................................................................
Supplementary Question 3: Firm Disclosures ..........................................
14
16
16
17
20
22
26
27
CONCLUSIONS .........................................................................................
32
REFERENCES ..........................................................................................................
35
APPENDIX A ...........................................................................................................
42
APPENDIX B ...........................................................................................................
44
V
28
30
vi
Page
APPENDIX C ...........................................................................................................
47
VITA .........................................................................................................................
62
vii
LIST OF TABLES
TABLE
Page
1
Sample derivation and distribution ............................................................
49
2
Descriptive Statistics ..................................................................................
51
3
Change in deferred-revenue account upon SOP 97-2 implementation ......
52
4
Whether earnings informativeness declines upon SOP 97-2
implementation ..........................................................................................
55
Whether the market considers the SOP 97-2-created deferred revenues as
revenues or as current liability ...................................................................
56
Whether the stock prices decline when firms report increases in deferredrevenue accounts upon SOP 97-2 implementation ....................................
58
5
6
7
Effects of SOP 97-2 – representative disclosures from firms’ 10-K filings,
before and after SOP 97-2 implementation ................................................ 59
1
I. INTRODUCTION
Revenue typically represents the largest and most value-relevant item in firms’
financial statements (FASB 2002). However, as more and more firms adopt a
“relationship marketing” approach (Galbraith 2002, Ghosh and Balachander 2007) by
bundling multiple products and services, revenue-recognition challenges keep mounting
(FASB 2002). Consider a software firm that delivers its base software on the contract
signing date and delivers integration services, maintenance, and software upgrades at
different times over the next 24 months. How does this firm determine revenues at any
time before the completion of this contract? Such a problem challenges both financial
statement preparers and standard setters (FASB 2002) and has figured prominently in
FASB’s deliberations on creating a new revenue-recognition standard (FASB/IASB
2007).1
The challenges of creating revenue-recognition standards for firms that bundle
multiple products and services (the “multiple-element” firms) differ from those of other
firms in at least one significant aspect – estimating revenues for partial deliveries of
multiple-element firms requires greater subjective assessments. Consider the above
example. How does this firm allocate aggregate contract value to individual elements
that are not sold separately? And how does it decide whether a delivered element
provides functional value to a customer pending delivery of another element? Estimating
This dissertation follows the style of The Accounting Review.
1
FASB’s advisory council has identified this agenda item as FASB’s highest priority project (FASAC
2006). FASAC has also recommended a review of the software industry’s revenue-recognition rules.
2
revenues based on the concept of completion of the earnings process (SFAC No. 5)
necessitates such assessments.
Nevertheless, the current revenue-recognition standards impose strict objectivity
and verifiability conditions rather than rely on managers’ subjective assessments. The
current standards require that multiple-element firms should defer revenue recognition in
its entirety even when firms have 1) delivered some elements substantially, but not
completely; 2) delivered some elements completely, but cannot objectively apportion the
contract value to these elements; and 3) delivered some elements completely, but cannot
establish their functionality pending delivery of other elements. In this study, I examine
whether such strict objectivity and verifiability conditions reduce multiple-element
firms’ earnings informativeness.2 On the one hand, such conditions might improve the
earnings informativeness of firms that had previously exploited relatively obscure
verifiability conditions to front-load revenue recognition. On the other hand, this “all or
nothing” approach can force a firm to defer its entire revenues even when it has partially
or completely delivered elements of significant economic value. For example, the
current rules preclude Apple Inc. from recognizing any revenue on the day it sells an
iPhone.
To investigate the effects of the multiple-element attribute of current standards, I
examine the effects of implementing SOP 97-2, rather than those of implementing SAB
2
This question remains unexamined. The current standards (i.e., SOP 97-2 [AICPA 1996] for the software
industry and SAB 101[SEC 1999] for the other industries) affect firms’ revenue-recognition practices in
two distinct ways: (1) through the completion of product-delivery requirement and (2) through additional
compliance requirements for multiple-element contracts. Zhang (2005) examines the effects of the first
aspect in isolation and Altamuro, Beatty, and Weber (2005) examine the simultaneous effects of both
aspects. However, the distinct effect of the second aspect remains unexamined.
3
101. Although SEC enacted SAB 101 three years after AICPA enacted SOP 97-2 and the
revenue-recognition requirements of SAB 101 virtually coincide with those of SOP 972,3 I use the earlier standard for three reasons: first, SAB 101 codified the multipleelement rules of SOP 97-2 and also introduced delivery and earnings-completion
requirements to those industries that were not formally covered by these rules earlier.
Therefore, empirical effects of implementing SAB 101 could arise from reasons other
than its multiple-element attribute (Moffeit and Eikner 2003). In contrast, because SOP
91-1, which preceded SOP 97-2, already required software firms to comply with
delivery and earnings-completion based conditions, the empirical effects of
implementing SOP 97-2 would arise only from its multiple-element attribute
(Carmichael 1998).
Second, software firms exemplify multiple-element firms. At a minimum, almost
all software firms provide future updates, upgrades, and extended maintenance support
in addition to base software. A significant number of software firms also provide
additional services such as customization, integration, and training.
Third, because SAB 101 applies rules originally designed for the software
industry to other industries, some might argue that the SEC misapplied SOP 97-2 rules
(FASB 2002). Specifically, AICPA created SOP 97-2 to prevent software firms from
prematurely recognizing revenues from multiple-element contracts (Carmichael 1998).
Therefore, implementing SOP 97-2 might improve software firms’ earnings
3
Both SOP 97-2 and SAB 101 build upon the concepts of earnings process completion and revenue
realizability (SFAC No. 5). Both require compliance with four necessary conditions before firms can
recognize revenues: (1) a formal arrangement, (2) completion of delivery, (3) determinable fees, and (4)
revenue realizability.
4
informativeness; nevertheless, this premise remains unexamined. While Altamuro et al.
(2005) report a general decline in earnings informativeness upon implementing SAB
101, their sample of SAB 101 affected firms is unlikely to include software firms
because SOP 97-2 preceded SAB 101.
I use three different research designs to examine my research question. First, I
use a before-and-after design to examine whether the market returns more strongly
correspond with pre-SOP 97-2 earnings or with SOP 97-2 implementation year’s
earnings. I find that the market returns more strongly correspond with former than with
latter, which suggests that implementing SOP 97-2 reduces earnings-returns association.
My second research design controls for the possibility that the above intertemporal decline in earnings-returns association occurs due to factors other than the SOP
97-2 implementation. I capitalize on balance sheet changes that result from
implementing stringent SOP 97-2 rules – the additional revenue deferrals, which I call
SOP 97-2-created deferrals. Under SOP 97-2 provisions, firms report these deferrals as
unearned or deferred revenues (i.e., as a current liability). However, under the prior
regime, firms would not have deferred these additional amounts and would have
recognized them as revenues. Therefore, I examine: Does the market price SOP 97-2created deferrals as revenues (consistent with the prior regime) or as current liabilities
(consistent with SOP 97-2)? Unlike the before-and-after design, this design enables me
to directly compare the two accounting methods using only the implementation year
data. I find that the market prices the SOP 97-2-created deferrals as revenues and not as
5
current liabilities. This finding supports the notion that the market prices earnings as if
the prior accounting regime still prevails and SOP 97-2 had not been implemented.
I provide further evidence on the above notion by using a third research design. I
examine whether the proforma earnings (i.e., the earnings that would have been reported
had SOP 97-2 not been implemented) are more informative than the reported earnings. I
calculate proforma revenues and earnings by adding back SOP 97-2-created deferrals. I
find that the proforma earnings more strongly relate with market returns than do the
reported earnings.
I contribute to the literature by documenting that SOP 97-2 reduces earnings
informativeness of the software firms. My findings suggest that a standard that requires
highly objective and verifiable support for multiple-element firms’ revenue assessments
reduces earnings informativeness, perhaps because estimating revenues for such firms
necessitates subjective assessments. My findings also dispel the notion that
implementing the current general standard (i.e., SAB 101) in non-software industries
reduces earnings informativess because the SEC misapplied SOP 97-2 rules to such nonintended industries.
My findings might interest the FASB in its revenue-recognition standard project
(FASB/IASB 2007). Standard setters’ decisions often involve trade-offs between
relevance and reliability. AICPA created SOP 97-2 to improve multiple-element firms’
earnings reliability by requiring such firms to defer revenues when they cannot
objectively and verifiably determine the extent of completion of the earnings process.
6
My results suggest that a more timely recognition of such deferred amounts might
convey more useful information to financial statement users.
The rest of the paper proceeds as follows. Section II provides background
information on SOP 97-2, Section III develops the hypotheses, Section IV describes the
sample, the research design, and the results, and Section V presents some concluding
remarks.
7
II. BACKGROUND INFORMATION ON SOP 97-2
Appendix A describes the changes in firms’ revenue recognition practices due to
the implementation of SOP 91-1, SOP 97-2, and SAB 101. Each of these three revenue
recognition standards builds upon the concept of completion of the earnings process
(SFAC No. 5). However, only SOP 97-2 provides the ideal setting to examine the
exclusive effects of the multiple-element attribute of current standards, because: 1) SOP
91-1 did not include the current multiple-element rules; and 2) most of the SAB 101
implementation effects arise from delivery-based revenue-recognition requirements,
rather than from its multiple-element rules. For example, the most common changes
from implementing SAB 101 include: 1) recognizing revenues for FOB goods only after
a firm transfers title of goods, 2) recognizing research fees over the contracted period
rather than upfront, and 3) recognizing revenues on layaways only after the customer
takes possession of the goods. Because of these changes, the pharmaceutical industry is
among the most SAB 101 affected industries (Moffeit and Eikner 2003; Altamuro et al.
2005), even though pharmaceutical firms do not assume significant multiple-element
obligations.
In this section, I describe the circumstances that led to creation of SOP 97-2 and
the reasons for its likely effects. In the mid 1990s, when many software firms were
growing rapidly and reporting losses, investors began to rely on revenues rather than on
earnings to value these firms (e.g., Damodaran 2002; Bowen, Davis, and Rajgopal
2002). Stock prices frequently reacted even if a firm met its earnings target, but missed
its revenue-growth target (Ertimur and Livnat 2003). Consequently, some software firms
8
began to bend accounting rules to report higher revenues. Notably, some software firms
would recognize entire revenues from a multiple-element contract even before they
delivered all the contracted elements (Carmichael 1998). The mid-1990s also witnessed
numerous revenue-related accounting restatements (e.g., GAO 2003, Palmrose, Scholz,
and Wahlen 2004). The SEC initiated enforcement actions against firms that recognized
revenues prematurely (e.g., COSO 1999). As revenue recognition became the SEC’s
priority, it requested that AICPA create stringent revenue recognition rules for the
software industry.
SOP 91-1, which specified the then applicable revenue-recognition rules, did not
provide explicit guidance on when to recognize revenues for undelivered elements in a
multiple-element contract. Specifically, the SOP 91-1 rules did not provide clear
guidance on: 1) how to estimate “significance” of undelivered elements; and 2) how to
account for software upgrades, which firms provide on a “when-and-if-available” basis.
Since SOP 91-1 had become effective, most software firms had begun bundling more
products and services in the same revenue arrangements by capitalizing on progress in
internet and distributed computing technologies (Choi and Whinston 1999; Fuerderer et
al. 1999). At a minimum, most software firms had begun providing extended
maintenance support through online updates and upgrades.
AICPA created SOP 97-2 to require software firms to defer recognizing a portion
of revenues corresponding to their undelivered elements (MacDonald 1996). In order to
achieve such revenue deferrals, AICPA added more elaborate and stringent conditions
before firms could recognize revenues from multiple-element contracts. For example,
9
SOP 97-2 added a more stringent interpretation of the delivery condition and introduced
a new vendor specific objective evidence (VSOE) requirement. The delivery condition
requires that all other elements essential for the functioning of the delivered element
should also have been delivered. The VSOE condition requires that a firm should first
establish the fair value of each element based solely on the firm’s own pricing records.4
Importantly, if a firm cannot establish fair value of just one element, the firm might not
be able to objectively apportion the aggregate contract value to any other element. This
might cause deferral of entire revenues from the contract (Carmichael 1998).
The importance of these conditions can be illustrated using the example of the
software firm described earlier. In order to recognize revenue for the base software on its
delivery date, the firm should 1) establish each element’s fair value based on the firm’s
own prior transactions and 2) ensure that the base software provides functional without
additional services. If the firm cannot meet either of these two conditions, it should defer
recognizing revenues corresponding to base software until (1) it meets both conditions;
or (2) it delivers all other elements.5
When AICPA issued an exposure draft of SOP 97-2 and invited comments,
several respondents protested against VSOE and delivery conditions (see Appendix B
4
The criteria used in practice to establish VSOE are stringent. A firm should provide evidence from at
least 30 prior randomly selected transactions and 85% of such transactions should have been priced within
15% of the median price (Sondhi 2006). This requirement is excessively stringent for firms that customize
products, have high technological obsolescence rates, or use dynamic pricing.
5
Despite customers’ outcry, Apple Inc. recently asked its customers to pay a nominal fee of $ 1.99 to
download a software patch for the computers it had sold earlier. Otherwise, its auditors would question its
policy of not deferring a portion of revenues for an undelivered element. Earlier, Apple tried defending its
accounting choice by claiming that it could not estimate fair values of future patches at the time of the
original sale. However, in such situations, the current standard requires firms to defer entire revenues from
the contract (Reily 2007).
10
for typical responses). For example, PWC stated, “VSOE of fair value will lead to
deferral of all revenues, even in situations where software products having clear value
and immediate utility to customer have been delivered.” An academician felt that
requiring a firm to defer recognition of entire revenue from a multiple-element contract
only because the firm could not meet just one criterion for just one element, amounts to
“stopping all traffic on a freeway because one car is broken down.” Lucent said:
“services compose between 1% and 15% of total revenue… the company would be
forced to defer 100% of revenue until 100% of services were performed.” Massachusetts
Society of CPAs called this an “all or nothing” approach.
11
III. HYPOTHESIS DEVELOPMENT
As described above, in 1996, AICPA enacted SOP 97-2 to prevent multipleelement software firms from recognizing revenues prematurely and to require them to
defer recognizing a portion of revenues corresponding to their undelivered elements. In
other words, SOP 97-2 requires firms to follow an element-by-element approach, by
recognizing revenues for an element only after objectively and verifiably determining its
completion of earnings process. This might improve earnings informativeness of firms
that took advantage of relatively obscure rules of SOP 91-1 to prematurely recognize
revenues.
Nevertheless, as described earlier, SOP 97-2’s requirements characterize an “all
or nothing” approach. This approach may force firms to defer revenue recognition in its
entirety even when firms have partially or completely delivered elements of significant
economic value. Furthermore, the rules of SOP 97-2 are so elaborate and comprehensive
that they border on audit standard setting (Carmichael 1998).6 These elaborate rules may
reduce accounting discretion that managers use to communicate value-relevant
information (e.g., Healy and Wahlen 1999, Fields, Lys, and Vincent 2001). Indeed,
Altamuro et al. (2005) find that implementing SOP 97-2 rules in a broader industry
setting reduces earnings informativeness. I examine whether their finding holds in the
software industry, that is: 1) in the multiple-element context and 2) in the industry for
6
The underlying rules of SOP 97-2 are simple and intuitive. Nevertheless, their application to a variety of
situations requires long and complex guidance. For example, KPMG’s and PWC’s technical guidance to
assist software firms in implementing SOP 97-2 rules run 332 pages (KPMG 2005) and 500 pages (PWC
2005) long.
12
which these rules were originally created. I test the following hypothesis using market
association tests:7
H1: Implementing SOP 97-2 leads to a decline in association between earnings
and market returns.
I examine the above question using a before-and-after design. However, this
design suffers from a limitation: inter-temporal changes could arise from factors other
than those due to the SOP 97-2 implementation.8 Business practices of high-tech firms
have been highly susceptible to changes since the mid-1990s (e.g., McAfee and
Brynjolfsson, 2007). In order to control for any inter-temporal effects, I directly examine
whether the market prices the implementation year’s financial statement components
consistent with the pre-SOP 97-2 or the post-SOP 97-2 method.
Implementing SOP 97-2 leads to additional revenue deferrals (i.e., the SOP 97-2created deferrals) that firms report as current liabilities. However, based on the earlier
accounting method, firms would not have deferred these additional amounts and would
have recognized them as revenues. Accordingly, I examine whether the market considers
SOP 97-2-created deferrals as revenues or as current liabilities, using the following
hypothesis.
7
Beaver (1972) suggests that “(the accounting) method which is more highly impounded (in securities
prices) ought to be the method reported in financial statements.” Lev and Ohlson (1982) state that
providing information for valuation should be accounting’s “desirable” property. Kothari (2001) defines
earnings informativeness as the association between accounting earnings and market returns. Barth,
Beaver, and Landsman (2001) suggest that examining how well accounting numbers relate with prices
provides “fruitful insights” to standard setters.
8
For example, Zhang (2005) finds that passage of time improved earnings informativeness of the nonSOP 91-1-affected control group (76% of her software-firms sample). To control for inter-temporal
effects, both Zhang (2005) and Altamuro et al. (2005) use non-affected firm samples as control groups.
13
H2: The market prices SOP 97-2-created deferred revenues more similarly to
how it prices revenues than to how it prices current liabilities.
If the above test indicates that the market prices SOP 97-2-created deferrals as
revenues and not as current liabilities, it might indicate that the market prices earnings as
if SOP 97-2 had not been implemented. I directly examine this notion. First, I calculate
proforma revenues by adding back SOP 97-2-created deferrals because absent SOP 97-2,
these amounts would not have been deferred. Then, I assume that the software firms
report costs that do not match revenues (Morris 1992, Zhang 2005, Mulford 2006).9
Accordingly, I calculate proforma earnings by adding back SOP 97-2-created deferrals.
Next, I examine whether the market returns more strongly correspond with proforma
earnings (consistent with pre-SOP 97-2 method) than with reported earnings (consistent
with SOP 97-2). I use the following hypothesis:
H3: Proforma earnings are more strongly associated with market returns than
are reported earnings.
9
SFAS 86 (FASB 1985) allows a firm to capitalize its software development costs after it establishes
software’ technological feasibility. Nevertheless, Mulford (2006) finds that on average, software firms
expense more than 90% of their development costs in the period in which these costs are incurred. For
example, despite deferring iPhone’s sales revenues, Apple expenses engineering, sales, and marketing
costs as they are incurred.
14
IV. SAMPLE SELECTION, RESEARCH DESIGN AND RESULTS
Sample Selection
SOP 97-2 applies to all firms whose products contain a significant software
component. Such firms belong to different industries and not just the software industry.
However, to select my sample, I focused on firms in the software industry (having SIC
codes beginning with 737) because SOP 97-2 definitely applies to these firms. Within
these firms, I focused on pre-packaged software firms (SIC code 7372) and integrated
software and services firms (SIC code 7373), which not only comprise more than 90% of
all firms in the software industry, but routinely use multiple-element contracts. The
other smaller software sub-industries, which comprise the remaining 10% of software
firms, might include pure services firms and might not reflect characteristics of multipleelement firms.
I derived my sample from the Compustat database using the sample selection
procedure described in Table 1. For testing H1, I assumed that firms implemented SOP
97-2 rules in the same years as AICPA required them to do so (i.e., the “prescribed
implementation years”).10 To the extent this assumption is violated, it would bias against
my finding significant results. I needed data on changes in stock prices, sales, assets, and
earnings for the following two years: the year prior to the implementation year and the
implementation year. Therefore, I retained 408 firms with SIC codes 7372 and 7373 with
10
Compustat defines fiscal year based on the month of May. However, SOP 97-2 defines prescribed
implementation year based on a December cut-off. For firms with fiscal years ending in June through
November, the prescribed implementation year corresponds to Compustat fiscal year 1999, and 1998 for
the others.
15
valid data for each of the above variables for each of the three years ending with the
implementation year.
To test H2 and H3, I hand-collected data on deferred-revenue accounts for the
following four years: a) two years prior to the implementation year, to determine the
“normal” level of firms’ deferred-revenue accounts; b) the implementation year; and c)
the year after the implementation year. The SOP 97-2 implementation years could
correspond to Compustat fiscal years 1998 or 1999, depending on firms’ fiscal year-end
months. Hence, I retained 423 firms with SIC codes 7372 and 7373 with revenue and
assets data available in Compustat for each of the fiscal years 1996-2000 (i.e., 1998−2 to
1999+1).
For these 423 firms, I gathered data from the firms’ 10-K filings. I obtained data
on deferred-revenue accounts from the liability section of balance sheets or from
footnotes that provide details on smaller liabilities. I looked for words such as “deferred
revenue,” “unearned income,” “customer advances,” and “billings in excess of revenue.”
I further dropped 137 firms that did not provide details on deferred-revenue accounts.
This filter left me with 286 firms.
Next, I examined the “revenue recognition” section in the “significant accounting
policies” footnotes and also performed a keyword search for SOP 97-2, to determine the
fiscal years in which firms first implemented the SOP 97-2 rules.11 I could not determine
11
To ascertain the SOP 97-2 implementation year, I looked for explicit disclosures in firms’ 10-K filings.
If a firm didn’t explicitly disclose its implementation year, I examined changes in its revenue-recognition
policy.
16
implementation years for 35 firms, which left me with 251 firms that constitute my
sample for testing H2 and H3.
Normal and SOP 97-2-Created Deferred Revenue
I partitioned the implementation year’s deferred-revenue accounts into normal
deferred revenue and SOP 97-2-created deferred revenue components. Deferred-revenue
account refers to those cash receipts from customers, which pending conversion to
revenues, are reported as liabilities. I assumed that (1) implementing SOP 97-2 did not
affect firms’ cash operating cycles and (2) firms had incentives to recognize revenues
early. Therefore, any increase in the deferred-revenue account in the SOP 97-2
implementation year would likely reflect additional revenue-recognition restrictions. I
calculated each firm’s deferred revenue to sales ratios for the two years prior to the
implementation year. I called their average the normal ratio. Then, I multiplied each
firm’s normal ratio with its reported revenue in the implementation year. I called this
product the normal deferred revenue. This component represents the deferred-revenue
account if SOP 97-2 had not been implemented. Then, I subtracted the normal deferred
revenue from the reported deferred-revenue account, and called the residual the SOP 972-created deferred revenue. Hence, I used firms’ own prior operating ratios to partition
the deferred-revenue account into normal deferred revenue and SOP 97-2-created
deferred revenue components.
Descriptive Statistics
Tables 1 and 2 present the descriptive statistics of 408 firms (for testing H1) and
251 firms (for testing H2 and H3) in the implementation year. I discuss the descriptive
17
statistics of 251 firms; those of the 408 firms are similar. Table 1 Panel B shows that the
sample of 251 firms is comprised of 187 prepackaged software firms (75%) and 64
systems integrators (25%). Table 1 Panel C shows that only 4 firms (2%) implemented
SOP 97-2 earlier than the prescribed year; 219 (87%) implemented in the prescribed
year; and the remaining 28 (11%) implemented later than the prescribed year (many
firms restated their accounts due to late implementation). Table 1 Panel D shows that
72% of the sample firms implemented SOP 97-2 in 1998, 23% in 1999, and the
remaining 5% in 1997 and 2000. Table 2 Panel B shows that the sample firms had
average assets of $ 337 million (median $50 million). However, they had much higher
average market value of $ 1,966 million (median $ 85 million), which suggests that
investors had high growth expectations from these firms. Note the high coefficient of
variation (ratio of standard deviation to mean) that reaches 800% for some variables.
This shows large variation in the sample firms’ characteristics. The average age of the
firms was 12 years (median 11 years). However, firms’ revenues grew at an average rate
of 46% (median 18%). The average ROA was negative and more than half of the sample
firms incurred losses. The average SOP 97-2-created deferred revenue to assets
(revenue) ratio was 1.66% (2.31%).
Univariate Tests on Deferred-Revenue Accounts
I first confirmed that implementing SOP 97-2 significantly increased software
firms’ deferred revenues and that this increase did not reflect a time trend. I used
revenue (Compustat DATA 12) to deflate the deferred revenues because if the cash
operating cycle remains unchanged, the deferred revenues should increase
18
proportionately with revenues. Nevertheless, I also used alternate deflators of assets
(DATA 6) and market value (common shares outstanding [DATA25] × closing price
[DATA199]).
Table 3 Panel A shows that upon implementing SOP 97-2 rules, the deferred
revenue to revenue ratio increased for approximately two-thirds of the sample firms. In
value terms, this ratio increased by an average of 2.3% (calculated as Ratiot – Ratiot-1).
This change, on average, represents a 42% increase over the prior year’s ratio (calculated
as (Ratiot – Ratiot-1)/Ratiot-1). All these increases are statistically significant. I obtained
similar results using alternate deflators of assets and market value. I also estimated
changes in the year before and the year after the SOP 97-2 implementation and found
that they are not significant. Therefore, I conclude that the increases in the deferredrevenue accounts in the implementation year do not reflect a time trend.
Systems integrators (e.g., IBM) are more likely to defer additional revenues upon
SOP 97-2 implementation than the prepackaged software firms (e.g., Microsoft).
Systems integrators provide multiple additional services, such as customization,
integration, training, and maintenance, along with their base software. Their base
software typically doesn’t function without customization and integration services.
Therefore, they face a greater difficulty in meeting the delivery condition in the early
phases of a project. Moreover, because they customize their software to buyers’
specifications, they lack a history of substantially similar prior transactions.
Accordingly, they face a greater difficulty in establishing VSOE.
19
Table 3 Panel B shows that implementing SOP 97-2 rules increased deferredrevenue to revenue ratios of both prepackaged software firms and systems integrators.
For prepackaged software firms, this ratio increased by 2.1% from the earlier level of
12.1%. This change, on average, represents a 36% increase over the pre-implementation
year ratio. For systems integrators, this ratio increased by 3.0% from the earlier level of
7.0%. This change, on average, represents a 61% increase over the pre-implementation
year ratio. This percent increase for the systems integrators exceeds the percent increase
for the pre-packaged software firms, which suggests that the greater the difficulty in
estimating the earnings completion process, the greater the increase in deferred-revenue
accounts upon SOP 97-2 implementation.
Discussion - univariate tests on deferred-revenue accounts
The magnitude of increase in deferred-revenue accounts is not as dramatic as
some respondents expected. The average increase in deferred-revenue accounts amounts
to 2.3% of revenues. This magnitude does not support some respondents’ expectation of
large revenue deferrals (e.g., read PWC’s and Lucent’s comments in Appendix B). One
possible reason for this less consequential effect could be that instead of deferring entire
revenues, firms started using construction accounting rules (SOP 81-1 [AICPA 1981])
more often. These rules allow firms to recognize revenues ratably proportional to
passage of time or to provision of ancillary services. For example, Apple Inc. uses this
method to reconize its iPhone sales revenues ratably over the service period. Therefore,
the difference between the pre-SOP 97-2 method and the post-SOP 97-2 method could
20
reflect: (i) the difference between SOP 91-1 and SOP 97-2 rules; and (ii) a greater use of
construction accounting.
Nonetheless, recognizing revenues ratably based on construction accounting does
not reflect the discrete value-delivery attribute of the software industry. For example, a
software firm does not continually deliver its base software, which is typically the most
valuable item in its multiple-element contract. It is fully uploaded all at once. Similarly,
Apple Inc. delivers its iPhone in one discrete transaction. This discrete value-delivery
attribute also manifests in the software industry’s milestone-based billing practices.
Consequently, a mismatch between timeliness of value-delivery events and revenuerecognition might adversely affect multiple-element firms’ earnings informativenes.
Testing Hypothesis H1
To investigate SOP 97-2’s effects on earnings informativeness, I examined
changes in the earnings response coefficient (ERC) and the revenue response coefficient
(RRC) in the prescribed implementation year relative to those in the prior year (Kothari
2001, Altamuro et al. 2005, Zhang 2005). I regressed excess buy-and-hold returns over
fiscal year (measured by percent change in end-of-the-year stock prices [DATA199]
minus risk-free return [CRSP RF]) on changes in revenues (DATA 12) and net incomes
(DATA 172). I deflated the latter two variables by beginning assets (DATA 6). I
controlled for firm size (log of assets), Fama-French factors (i.e., excess market return
[CRSP VWRETD − CRSP RF], high minus low growth [CRSP HML], small minus big
size [CRSP SMB], and momentum [CRSP UMD]), and sub-industry and year fixed
effects. I used the following regression:
21
Excess_BHRit = β01 + β02 × After
+ β11 × ∆NetIncomeit + β12 × After × ∆NetIncomeit
+ β21 × ∆Revenueit
+ β22 × After × ∆Revenueit
+ Σ βs1 × Controlsit
+ Σ βs2 × After × Controlsit + εit
(1)
where
∆ = one-year change;
After = dummy variables set to 1 for observations in the prescribed
implementation year, and 0 for the year prior to that year;
i = 1 to 408 firms; and
t = the prescribed implementation year and the year prior to that year.
A negative β12 and β22 would indicate that ERC and RRC, respectively, decline
due to implementation of SOP 97-2 rules.
Results for hypothesis H1
Table 4 Panel A (equation 1) shows that β12 is significantly negative. This
indicates that ERC declined when firms implemented the SOP 97-2 rules. Furthermore,
the summation of fixed effect and the interaction effects (β02 + β12) of implementing SOP
97-2 is significantly negative. This result suggests that implementing SOP 97-2 rules
deteriorates earnings informativeness in the software industry. However, β22 is not
significantly negative.
These results are consistent with Altamuro et al. (2006) and are contrary to the
notion that implementing SAB 101 guidelines (i.e., the SOP 97-2 rules) deteriorates
earnings informativeness because they have been misapplied in a set of non-intended
22
industries. On the contrary, these results support Healy and Wahlen (1999) and Fields,
Lys, and Vincent (2001). Note the significantly positive coefficient on revenues despite
controlling for net income. This indicates that for growth firms, revenues provide
incremental value-relevant information beyond the information provided by earnings.
Testing Hypothesis H2
To test H2, I examined whether the market prices SOP 97-2-created deferred
revenues more similarly to how it prices the revenues than to how it prices normal
deferred revenues. I regressed excess buy-and-hold returns for the SOP 97-2
implementation year on SOP 97-2-created deferred revenues and changes in normal
deferred revenues and reported revenues. I controlled for changes in net incomes and
deflated all dollar-denominated variables by beginning assets. I also controlled for firm
size (log of assets) and firm age, which I estimated using incorporation-year data
obtained from the firms’ 10K filings. In addition, I controlled for Fama-French factors,
and sub-industry fixed effects and year fixed effects. I used the following regression:
Excess_BHRit = β0 + β1 × ∆NormalDeferredRevenuet
+ β2 × SOP97-2CreatedDeferredRevenueit
+ β3 × ∆Revenueit + β4 × ∆NetIncomeit
+ Σ βs × Controlsit + εit
where
∆ = one-year change;
i = 1 to 251 firms; and
t = the SOP 97-2 implementation year.
(2)
23
A positive β2 that significantly differs from β1 but not from β3 might indicate that
the market considers SOP 97-2-created deferred revenues as revenues and not as current
liability.
Upon implementing SOP 97-2, the subsequent year’s earnings would include the
effects of the implementation year’s SOP 97-2-created deferred revenues. However,
based on the pre-SOP 97-2 method, these revenues would not be deferred and the
subsequent year’s earnings would not include the effects of these deferrals. If the market
considers SOP 97-2-created deferred revenues as revenues, the market would price the
subsequent years’ earnings as if these earnings did not include the effects of SOP 97-2created deferrals. In other words, the market would price subsequent year’s earnings as if
by subtracting the portion of earnings that resulted from late recognition of the SOP 972-created deferred revenues. I examined this notion using the following test.
Let t represent the implementation year and let t+1 represent the subsequent year.
Typically, returns in year t+1 should be associated with changes in earnings as follows:
*
Rit +1 = β1 + β 20 × ∆Eit +1 + ε it +1
(3)
where, Rt+1 is the return, and ∆E*t+1 is the unobserved change in earnings, absent any
SOP 97-2-created deferral. Based on the pre-SOP 97-2 method, ∆Et+1, which is
observed, overstates ∆E*t+1 due to the implementation year’s SOP 97-2-created deferred
revenues (SOPCDRt), as follows:
*
∆Eit +1 = ∆Eit +1 + Margin × SOPCDRit
(4)
where margin represents net income to revenue ratio. Accordingly, the unobserved
change in earnings, based on the pre-SOP 97-2 method, would be as follows:
24
*
∆Eit +1 = ∆Eit +1 − Margin × SOPCDRit
(5)
Putting this in equation (3) gives:
Rit +1 = β1 + β 20 × (∆Eit +1 − Margin × SOPCDRit ) + ε it +1
(6)
Rit +1 = β1 − β 2 × SOPCDRit + β 20 × ∆Eit +1 + ε it +1
(7)
Therefore, controlling for changes in reported earnings in year t+1, if the
association between returns in year t+1 and SOPCDRt (i.e., β2 in equation 7) is negative,
it would indicate that the market prices the subsequent years’ earnings as if by
“adjusting” for the portion of earnings that result from late recognition of the SOPCDR.
Accordingly, I tested H2 based on combined tests of equation (1) and equation (8)
below:
Excess_BHRit+1 = β0 + β1 × ∆NormalDeferredRevenuet
+ β2 × SOP97-2CreatedDeferredRevenueit
+ β3 × ∆Revenueit+1 + β4 × ∆NetIncomeit+1
+ Σ βs × Controlsit+1 + εit+1
(8)
where
∆ = one-year change;
i = 1 to 251 firms;
t = the SOP 97-2 implementation year; and
t+1 = the year subsequent to the SOP 97-2 implementation year.
Results for hypothesis H2
The first column of Table 5 Panel A (equation 2) shows that the
contemporaneous returns are not significantly associated with increases in deferred-
25
revenue accounts. This could be because, on one hand, increases in deferred-revenue
accounts signal future revenues; on the other, they represent increases in current
liabilities. However, the results in the second column show that disaggregating the
deferred-revenue account into SOP 97-2-created deferred revenues and normal deferred
revenues results in a positive coefficient on SOP 97-2-created deferred revenues.
Importantly, this coefficient significantly differs from the coefficient on normal deferred
revenues, but not from the coefficient on reported revenues. This result suggests that the
market considers SOP 97-2-created deferred revenues as revenues and not as normal
deferred revenues.
The first two columns of Table 5 Panel B (equation 8) show that the subsequent
year’s returns are unrelated to the implementation year’s aggregate deferred revenues as
well as to its components (normal deferred revenue and SOP 97-2-created deferred
revenues). However, the third column shows that controlling for the subsequent year’s
reported revenues and net income, the coefficient on implementation year’s SOP 97-2created deferred revenues becomes negative. This negative sign suggests that the market
prices the subsequent-year earnings as if by subtracting the portion of earnings that
results from prior year’s SOP 97-2-created deferred revenues. Therefore, both the
implementation and the subsequent years’ results support the premise that the pre-SOP
97-2 accounting method more strongly corresponds with market returns.
I also regressed the market value at the end of the implementation year on
disaggregated balance sheet items. I deflated all dollar-denominated variables by the
number of outstanding shares. Untabulated results show that the coefficients on assets
26
and liabilities are predictably positive and negative, respectively. Moreover, the
coefficient on normal deferred revenue is predictably negative and does not differ
significantly from that on current liabilities. On the contrary, the coefficient on SOP 972-created deferred revenues is positive and significantly differs from that on current
liabilities. Therefore, I obtained similar results using the balance-sheet items as those
obtained using the income-statement items.
Testing Hypothesis H3
To test H3, I examined whether a regression specification using proforma
earnings more strongly corresponds with market returns than does the specification using
reported earnings. I calculate proforma revenues and proforma earnings by adding back
SOP 97-2-created deferred revenues. I used the following regression:
Excess_BHRit = β0 + β1 × ∆NormalDeferredRevenuet
+ β2 × ∆ProformaRevenueit
+ β4 × ∆ProformaNetIncomeit
+ Σ βs × Controlsit + εit
(9)
where
∆ = one-year change;
∆ProformaRevenue = ∆Revenue + SOP 97-2-created deferred revenues;
∆ProformaNetIncome = ∆NetIncome + SOP 97-2-created deferred revenues;
i = 1 to 251 firms;
t = the SOP 97-2 implementation year.
27
I used the same control factors as described in equation 2. I examined whether
adjusted R-square of equation 9 exceeds that of equation 2.
Results for hypothesis H3
The third column of Table 5 Panel A shows that using proforma accounting
numbers results in a better specified model. Vuong test shows that the adjusted R-square
of specification using the proforma numbers exceeds that of specification using the
reported numbers. This suggests that if SOP 97-2 had not been implemented, the
accounting numbers would more strongly correspond with market returns than do the
reported numbers.
The results of the three hypotheses, using three different research designs,
suggest that the numbers calculated using pre-SOP 97-2 revenue-recognition rules more
strongly correspond with market returns than do those using SOP 97-2.12 Therefore, my
study complements Altamuro et al. (2005) by documenting that implementing SOP 97-2
reduces earnings informativeness in the software industry, the industry for which these
rules were originally created. More importantly, my results suggest that implementing
the current revenue-recognition standard reduces earnings informativeness of multipleelement firms.
Supplementary Question 1: Naïve Reaction
Did implementing SOP 97-2 cause any adverse capital market consequences?
Some market participants feared that reporting lower earnings (upon implementing SOP
12
Alternatively, these results could indicate that investors prefer to compare financial statements, which are
prepared using the same accounting rules (Sunder 1973). This explanation however, is less likely because
I measure returns before earnings are announced.
28
97-2) could adversely affect firms’ financial position and fund-raising abilities.13
Effectively, they feared that prices might not accurately reflect the mechanical earningsdecreasing effects of SOP 97-2 -created deferrals. Prior studies (e.g., Lev and Ohlson
1982, Hirshleifer and Teoh 2003) concluded that the naïve reaction or functional fixation
hypothesis cannot be entirely dismissed.
The difference between reported earnings and earnings based on the pre-SOP 972 method would arise primarily because of SOP 97-2-created deferrals. The above
market participants’ concerns would be valid if the market prices earnings as they are
reported and correspondingly, prices SOP 97-2-created deferrals as current liabilities.
However, my results suggest that the market prices the SOP 97-2-created deferrals as
revenues and not as current liabilities. Therefore, I do not find statistical support for the
notion that reporting lower earnings upon implementing SOP 97-2 leads to adverse
capital market consequences.
Supplementary Question 2: Aggressive Revenue-Recognition Before SOP 97-2 Was
Enacted
Were the software firms systematically recognizing revenues prematurely before
AICPA enacted SOP 97-2? AICPA formulated stringent revenue-recognition rules based
on this premise (Carmichael 1998). Accordingly, implementing SOP 97-2 rules would
“correct” those software firms’ revenue-recognition timeliness that had been recognizing
revenues prematurely. Such “correction” would result in additional revenue deferrals;
13
For example, Haushahn Systems in its response letter states, “…would have a substantial negative
impact on our financial position.” Ascential Software Corp. states in its 1998 10-K filing, “…will have a
material adverse affect on the Company's revenues, gross margins and operating results. As a result, future
capital raising efforts may also be adversely affected.”
29
the higher the correction, the greater the additional revenue deferrals. In other words,
reporting increases in deferred-revenues accounts might signal firms’ prior aggressive
revenue-recognition practices. Prior studies have documented that stock prices decline
when firms reveal use of aggressive accounting practices.14 I would find support for this
premise if (1) stock-prices decline when firms report increases in deferred-revenues
accounts; or (2) the magnitude of these price-declines increases with rules-created
deferrals.
I calculated 7-day cumulative returns surrounding the days when firms reported
the details on deferred-revenue accounts: the earnings announcement dates (Compustat
RDQE of 4th quarter) and the 10-K filing dates (hand collected). Then, I used the
following regression:
7-dayReturnit = β01 + β02× DummyIncreaset
+ β1 × ∆NormalDeferredRevenueit
+ β21 × SOP97-2CreatedDeferredRevenueit
+ β22 × DummyIncreaset × SOP97-2CreatedDeferredRevenueit
+ β3 × ∆Revenueit + β4 × ∆NetIncomeit
+ β5 × 7-dayMarketReturnt + Σ βs× Controlsit + εit
(10)
Where
∆ = one-year change;
DummyIncrease = dummy variables set to 1 for observations with positive SOP
97-2-created deferred revenues, and 0 otherwise;
14
For example, when firms announce: (1) SEC’s AAER investigations (Feroz, Park, and Pastena 1991);
and (2) accounting restatements (Palmrose, Richardson, and Scholz 2004).
30
i = 1 to 212 firms; and
t = the SOP 97-2 implementation year
I controlled for market returns, size, age, and firm and sub-industry fixedeffects.15 A negative β02 or β22 would support AICPA’s premise. A negative β02 might
indicate that prices decline when firms reveal increases in deferred-revenue accounts.
And a negative β22 might indicate that these price-declines increase with magnitude of
increases in deferred-revenue accounts.
The results, presented in Table 6, show that neither β02 nor β22 is significantly
negative. Therefore, I do not find statistical support for the premise that firms
systematically recognized revenues prematurely before AICPA enacted SOP 97-2.
Supplementary Question 3: Firm Disclosures
How did investors obtain information to distinguish SOP 97-2-created deferred
revenues from normal deferred revenues? I examined one possible source of investor
information – firms’ narrative disclosures in their 10-K filings. SEC rules require firms
to discuss the effects of implementing a new accounting rule, such as SOP 97-2, both
before and after its implementation: (1) before its implementation, SAB 74 (SEC 1987)
requires firms to discuss the likely changes; and (2) after its implementation, MD&A
rules (SEC 1989) require firms to discuss the material changes in line-items, especially if
these changes relate to revenue items. Moreover, academic research predicts that firms
will increase their narrative disclosures when accounting rules reduce firms’ accounting
discretion (Dye 1985, Kasznik 1996).
15
Ideally, I should use analysts’ forecasts to estimate revenue and earnings surprises. However, I could not
find analysts’ forecasts in the IBES database for most of my sample firms.
31
Table 7 Panel A presents a sample of the before disclosures. In their 10-K filings,
most firms described that they planned to implement SOP 97-2 in their next fiscal year.
However, they did not provide many details about forthcoming accounting changes.
Most firms stated that they did not expect significant changes from implementing the
SOP 97-2 rules. Table 7 Panel B presents a sample of the after disclosures. Similar to
the before disclosures, most firms did not provide many details on SOP 97-2 effects.
Firms typically stated: “No material effect to the company’s operation or financial
position taken as a whole,” (e.g., Nyfix Inc.). Very few firms quantified the effects of
implementing SOP 97-2 rules. Therefore, I conjecture that the market relied on nonfinancial sources to distinguish between the two deferred-revenue components. How the
market prices these two components differently remains a topic for future research.
32
V. CONCLUSIONS
Revenue is among the largest and most value-relevant items in firms’ financial
statements. However, it is not easy to determine revenues for firms that use the
“customer-centric” and “relationship marketing” approach (i.e., those firms that bundle
multiple products and services and deliver them over extended periods). As more and
more firms adopt a relationship-marketing approach, the complexity of revenuerecognition issues increases.
The dilemma for standard setters remains: What is the optimum level of
objectivity and verifiability requirements that would enhance the two important
“qualitative characteristics” of financial statements: relevance and reliability (FASB
2005)? Standard setters’ rules-making processes often involve trade-offs between these
two characteristics. On the one hand, if standard setters make rules too conservative, by
requiring firms to comply with highly objective and verifiable conditions, revenues may
not reflect economic performance in a timely manner (affects relevance). On the other
hand, if standard setters make the rules too lax, firms can use this freedom to recognize
revenues prematurely (affects reliability).
Standard setters would have faced a similar dilemma when they formulated SOP
97-2. In response to software firms’ perceived premature revenue-recognition practices,
AICPA increased objectivity and verifiability requirements before firms could recognize
revenues from multiple-element contracts. However, estimating whether a multipleelement firm has “earned” revenues routinely involves subjective estimates. For
example: 1) in allocating aggregate contract value to various elements that are not sold
33
separately and 2) in deciding whether an element provides functional value pending
delivery of another element. Nevertheless, SOP 97-2 requires a firm to defer recognizing
entire revenues when it cannot objectively and verifiably determine the extent of earned
revenues from a multiple-element contract.
I find that implementing SOP 97-2 increases revenue deferrals and reduces
software firms’ earnings informativeness. In other words, imposing highly objective and
verifiable conditions before firms can recognize revenues from multiple-element
contracts leads to deferral of revenues, which managers would have otherwise
recognized using subjective estimates. My finding suggests that recognizing such
deferred amounts in a timely manner conveys more useful information to financial
statement users.
My findings should be useful to FASB in its current deliberations (FASB/IASB
2007). FASB is evaluating alternative revenue-recognition approaches, which include:
(1) the measurement model (i.e., the fair-value approach) and (2) the allocation model
(i.e., the customer-consideration approach). The latter model is constructively similar to
the earnings-completion model. My study suggests that imposing highly objective
conditions for multiple-element revenue recognition could lead to economic effects
similar to those from SOP 97-2 implementation.
Admittedly, I have examined only one criterion – how accounting numbers
correspond with securities prices – among several that guide the standard setters’
decisions. Holthausen and Watts (2001) conclude that evaluating accounting rules using
only the relevance criterion does not adequately describe the rules-setting process.
34
Nevertheless, Barth et al. (2001) argue that share prices reflect accounting numbers, only
if the relevant information is measured and reflected reliably in the accounting numbers.
35
REFERENCES
Aboody, D., M. E. Barth and R. Kasznik. (2004). SFAS 123 stock-based compensation
expense and equity market values. The Accounting Review 79: 251-275.
Altamuro, J., A. L. Beatty, and J. P. Weber. 2005. The effects of accelerated revenue
recognition on earnings management and earnings informativeness: Evidence from
SEC staff accounting bulletin No. 101. The Accounting Review 40(2): 373-401.
American Institute of Certified Public Accountants (AICPA). 1981 Accounting for
Performance of Construction Type and Certain Production Type Contracts.
Statement of Position (SOP) 81-1. AICPA, New York.
American Institute of Certified Public Accountants (AICPA). 1991. Software Revenue
Recognition. Statement of Position (SOP) 91-1. AICPA, New York.
American Institute of Certified Public Accountants (AICPA). 1997. Software Revenue
Recognition. Statement of Position (SOP) 97-2. AICPA, New York.
Ball, R. J. 1972. Changes in accounting techniques and stock prices. Empirical research
in Accounting: Selected Studies. Journal of Accounting Research 10: 159-78.
Barth, M. E., W. H. Beaver, and W. R. Landsman. 2001. The relevance of the value
relevance literature for financial accounting standard setting: Another view. Journal
of Accounting and Economics 31: 77–104.
36
Beaver, W. H. 1972. The behavior of security prices and its implications for accounting
research (methods). Committee Reports. Supplement to The Accounting Review:
407-437.
Bowen, R. M., A. K. Davis, and S. Rajgopal. 2002. Determinants of revenue-reporting
practices for internet firms. Contemporary Accounting Research 19(4): 523-562.
Carmichael, D. R. 1998. Software revenue recognition under SOP 97-2. CPA Journal
68(7): 44-51.
Choi, S. Y., and A. B. Whinston. 1999. The future of e-commerce: Integrate and
customize. Computer 32(1): 133-134.
Committee of Sponsoring Organizations of the Treadway Commission (COSO). 1999.
Fraudulent financial reporting: 1987-1997 - An analysis of U.S. public companies.
Available at: http://www.coso.org/publications/FFR_1987_1997.PDF
Damodaran, A. 2002. Investment valuation: Tools and techniques for determining the
value of any asset. Second edition, Wiley. Hoboken, NJ.
Davis, K. A. 2002. The value relevance of revenue for internet firms: Does reporting
grossed-up or barter revenue make a difference? Journal of Accounting Research 40
(2): 445–477.
Dye, R. 1985. Strategic accounting choice and the effects of alternative financial
reporting requirements. Journal of Accounting Research 23(2): 544-574.
37
Ertimur, Y. and J. Livnat. 2003. Differential market reactions to revenue and expense
surprises. Review of Accounting Studies 8: 185-211.
Feroz, E.H., K. Park; and V.S. Pastena. 1991. The financial and market effects of the
SEC's accounting and auditing enforcement releases. Journal of Accounting
Research 29:107-148.
Fields, T., T. Lys, and L.Vincent. 2001, Empirical research on accounting choice.
Journal of Accounting & Economics 31, 255-307.
Financial Accounting Standards Advisory Council (FASAC). 2006. Results of the
Annual Survey on the Priorities of the FASB (October 2006).
Financial Accounting Standards Board (FASB). 1984. Concepts Statement No. 5.
Recognition and Measurement in Financial Statements of Business Enterprises
(December 1984).
Financial Accounting Standards Board (FASB). 1985. Financial Accounting Standards
No. 86. Accounting for the Costs of Computer Software to Be Sold, Leased, or
Otherwise Marketed. (August 1985).
Financial Accounting Standards Board (FASB). 2000. EITF00-21. Revenue
Arrangements with Multiple Deliverables (July 2000).
Financial Accounting Standards Board (FASB). 2002. Minutes of the May 15, 2002
board meeting—Proposal for a New Agenda Project on Issues Related to the
Recognition.
38
Financial Accounting Standards Board (FASB). 2005. Relevance and Reliability. The
FASB Report (February 28, 2005).
Financial Accounting Standards Board/ International Accounting Standards Board
(FASB/IASB). 2007. Minutes of the October 22, 2007 FASB-IASB joint board
meeting.
Fuerderer, R., A. Herrmann, and G, Wuebker. 1999. Optimal bundling: Marketing
strategies for improving economic performance. First edition, Springer. New York,
NY.
Galbraith, J. R. 2002. Organizing to deliver solutions. Organizational Dynamics 31(2):
194-206.
General Accounting Office (GAO). 2003. Financial Statement Restatements Database.
Report 03-395R. Government Printing Office. Washington D.C.
Ghosh, B. and S. Balachander. 2007. Competitive bundling and counter-bundling with
generalist and specialist firms. Management Science 53(1): 159-168.
Healy, P. M. and J. M. Wahlen. 1999. A review of the earnings management literature
and its implications for standard setting. Accounting Horizons (December): 365-383.
Hirshleifer, D., and Teoh, S. H. 2003. Limited attention, information disclosure, and
financial reporting. Journal of Accounting and Economics 36(1-3): 337-386.
Kasznik, R. 1996. Financial reporting discretion and corporate voluntary disclosure:
Evidence from the software industry. Working paper, Stanford University.
39
Kothari, S. P. 2001. Capital market research in accounting. Journal of Accounting and
Economics 31(1): 233-253.
Klynveld Peat Marwick Goerdeler (KPMG). 2005. Software Revenue Recognition. An
Analysis of SOP 97-2 and Related Guidance. Second edition. Available at:
www.us.kpmg.com/microsite/attachments/2005/SoftwareRevRecognitionBook2005.
pdf
Lev, B., and J. A. Ohlson. 1982. Market-based empirical research in accounting: A
review, interpretation, and extension. Journal of Accounting Research 20
(supplement): 249-322.
Lowell, M. J. 1994. Software revenue recognition. The CPA Journal Online (August).
Available at http://www.nysscpa.org/cpajournal/old/15703011.htm
MacDonald, E. 1996. Accounting group considers forcing software companies to defer
revenue. Wall Street Journal (November 11).
McAfee, A., and E. Brynjolfsson. 2007. Dog Eat Dog. The Wall Street Journal (April
28): R.10.
Moffeit, K. S., and A. E. Eikner. 2003. Implementation of SAB 10. CPA Journal Online
(January). Available at:
http://www.nysscpa.org/cpajournal/2003/0103/dept/d015603.htm
Morris, J., 1992. Software Industry Accounting. First edition, Wiley. Hoboken, NJ.
40
Mulford, C. W. 2006. Capitalization of software development costs – A survey of
accounting practices in the software industry. A monograph from Georgia Tech.
College of Management.
Osterland, A. 2000. Hard lessons. CFO Magazine (September).
Palmrose, Z. V., Richardson, V. J., Scholz, S. W., 2004. Determinants of market
reactions to restatement announcements. Journal of Accounting and Economics 37,
59-89.
Palmrose, Z. V., Scholz. S., and Wahlen, J. M. 2004. The circumstances and legal
consequences of non-GAAP reporting: Evidence from restatements. Contemporary
Accounting Research 21(1):139-190.
PricewaterhouseCoopers (PWC). 2005. Software Revenue Recognition. A user-friendly
guide for navigating through the many complexities. Fourth edition. Available from
Global Software Industry Leader, PWC.
Public Company Accounting Oversight Board (PCAOB). 2007. Report on the PCAOB'S
2004, 2005, and 2006 Inspections of Domestic Triennially Inspected Firms.
Available at: http://www.pcaobus.org/Inspections/Other/2007/1022_4010_Report.pdf
Reily, D. 2007. Apple Gets a Bruise by Blaming a $ 1.99 Fee on Accounting Rules. The
Wall Street Journal Eastern edition (Jan 20): B.3.
41
Securities and Exchange Commission (SEC). 1987. Staff Accounting Bulletin No. 74
(SAB 74), Disclosure of the Impact that Recently Issued Accounting Standards Will
Have on the Financial Statements of the Registrant when Adopted in a Future Period.
Securities and Exchange Commission (SEC). 1989. SEC Interpretation: Management's
Discussion and Analysis of Financial Condition and Results of Operations; Certain
Investment Company Disclosures (May 18).
Securities and Exchange Commission (SEC). 1999. Staff Accounting Bulletin No. 101
(SAB 101), Revenue Recognition in Financial Statements (December 3).
Securities and Exchange Commission (SEC). 2003. Staff Accounting Bulletin No. 104
(SAB 104), Revenue Recognition, Corrected Copy (December 17).
Sondhi, A. C. 2006. SOP 97-2: Current issues in VSOE accounting. Presentation to
CFOs available at CFO.com.
Sunder, S. 1973. Relationship between accounting changes and stock prices: Problems
of measurement and some empirical evidence. Journal of Accounting Research 11:
1-45.
Zhang, Y. 2005. Revenue recognition timing and attributes of reported revenue: The
case of software industry's implementation of SOP 91-1. Journal of Accounting and
Economics. 39(3): 535-561.
APPENDIX A
SIGNIFICANT CHANGES IN REVENUE-RECOGNITION PRACTICES UPON IMPLEMENTING SOP 91-1,
SOP 97-2 AND SAB 101*
Panel A: SOP 91-1 implementation in the software industry
Revenue recognition criteria
Before
After
Significant changes in industry practices
A 1983, survey results showed 1. Delivered or
• License with no vendor obligations: Recognize revenue upon delivery of the
performed
a substantial diversity in
software.
revenue recognition practices
2. Realizable
• License with insignificant vendor obligations: Accrue costs of insignificant
ranging from the very
obligations or defer a pro-rata portion of revenue and recognize it ratably as
conservative (on customer
the costs are incurred.
payment) to the decidedly
• License with significant vendor obligations: Recognize revenue according to
aggressive (immediately on
the contract accounting method or as a service transaction
contract signing). 15% of the
• Post-contract support: Recognize revenue over the period during which the
companies used contract
services are provided.
signing-date to recognize
entire revenues.
Panel B: SOP 97-2 implementation in the software industry
Revenue recognition criteria
Before
After
Significant changes in industry practices
1. Delivered or performed
1. Evidence of
• Element-by-element revenue recognition
arrangement
• If functionality of a delivered element requires delivery of another element,
2. Delivered or
then revenues for that element should be deferred. Otherwise, firm should
performed
recognize entire revenues ratably (SOP 81-1).
(expanded
• If the contract promises a “when-and-if available” element, such as an
definition)
unspecified upgrade, such an element creates a contingency that should be
accounted as a separate element.
2. Realizable
3. Realizable
• Firm should objectively and verifiably determine fair values of all elements
4. Determinable
based on its own prior transactions. Otherwise it should defer revenues in its
fees
entirety or recognize them ratably (SOP 81-1).
42
Panel C: SAB 101/SAB 104 implementation in the non-software industries
Revenue-recognition criteria
Before
After
Significant changes in industry practices ‡
Several industry1. Evidence of
• Licensing arrangements: Delivery and revenue recognition don't occur until
specific and general
arrangement
the term of the license begins.
standards
2. Delivered or
• Layaway programs: Sellers should not record revenue until the product is
performed†
delivered to the customer.
3. Realizable
• Upfront fees: Even if nonrefundable, these fees should be deferred and
4. Determinable fees†
recognized over the term of the agreement.
• Setup services: Should be recognized on a straight-line basis over the term of
†Multiple-element
the contract, even if most of the costs are incurred up front.
conditions later clarified
• Contingent rent: Income contingent on a factor other than time should be
through SAB 104 (SEC
recorded only when the contingency is resolved.
2000) and EITF00-21
• Rental income: Retailers can recognize only the rental income from leased or
(FASB 2002)
licensed departments, not that department's revenues.
‡92% of firms that changed their revenue-recognition practices did it to comply
with the new delivery condition.
*References: Lowell (1994), Carmichael (1998), Osterland (2000), and Moffeit and Eikner (2003).
43
APPENDIX B
REPRESENTATIVE COMMENTS RECEIVED BY AICPA ON THE DRAFT SOP 97-2*
Respondent/Issue
Quote from the comment letters
Firm should defer recognizing revenues corresponding to a delivered element if it cannot establish that element’s VSOE
Pricewaterhouse
We are troubled that in some multiple element arrangements the ED’s requirements for VSOE of fair value
will lead to deferral of all revenue, even in situations where software products having clear value and
immediate utility to customer have been delivered. At first blush, this seems unduly harsh.
BriarCliff College
Deferring revenue recognition for all other elements in a multiple element arrangement because one element in
bundle could not be fairly valued seems too conservative. This is like stopping all traffic on freeway because
one car is broken down.
Lucent Corp
We are troubled by the underlying assumption that all software companies can and do market all of their
products separately. There are some software companies which actively bundle their products and
services…..these companies may not be able to comply with VSOE evidence criteria.
Candle Corporation
Due to many pricing considerations and variables described above, the permutations and competitive pricing
makes it impractical to meet VSOE requirements
Massachusetts Society of CPAs
We believe that the concept of VSOE may in fact be quite difficult for software companies to actually apply in
their business….An “all or nothing” approach can significantly swing revenues…..
Firm cannot recognize revenue corresponding to a delivered element until it delivers all elements essential for the functioning of that element
Lucent Technologies
All of the criteria to recognize revenue may be met for the most significant elements, yet all of the revenue
would have to be deferred solely because… products sold in bundled package. E.g. services compose between
1% and 15% of total revenue… the company would be forced to defer 100% of revenue until 100% of services
were performed….
Pricewaterhouse
If for example, the vendor never sells training services on a discrete basis, the ED would seem to require
deferral of all revenue from the arrangement until the delivery of ancillary training services are completed. In
many circumstances, we do not believe such accounting would be appropriate.
44
Firm should allocate and defer recognizing revenues corresponding to specified upgrades. If the firm cannot determine the amount to allocate, it
should defer entire revenues from the contract
Hausahn Systems and Engineers
…Since we do not charge for upgrades, it is unclear whether we would have to allocate…
Arthur Anderson
SOP should be expanded with a discussion of how to determine VSOE of fair value of an upgrade right
Lucent Technologies
… We believe that it would be inappropriate to defer the entire amount of software revenue as a result of
specifies upgrade/enhancement that is not of significance to the transaction.
Respondent/Issue
Quote from the comment letters
Firm should defer revenues for the entire amount at risk due to acceptance/warranty/performance clause/right of return:
Arthur Anderson
We believe that routine acceptance clauses should not preclude revenue recognition at delivery.
I2 Technologies
In large enterprise software offerings where customer is spending million of dollars for mission
critical software, the customer often rightfully requires a warranty for a year or more. ….I believe
this should not impair revenue recognition
Implementing SOP 97-2 rules might lead to violation of matching principle
Lucent Technologies
Bulk of costs will already have been expensed…while revenue will not be recognized until all
elements are delivered….this could be a departure from matching principle and could be very
misleading
Hausahn Systems and Engineers
In our case, approximately 90% of baseline development costs are expensed as period cost. Deferral
of revenue will probably result in a lack of matching revenue and expense.
Firm should recognize license revenues ratably over the service period if it cannot establish VSOE for the license
Inso Corp
If sufficient VSOE does not exist to allocate the license fee to the separate elements, the entire
arrangement fee should be recognized ratably over the period during which PCS is expected to be
provided.
California Society of CPAs
To impose the “separate” pricing criteria is not realistic and will generally preclude revenue
recognition on initial delivery …..
Firms might find it difficult to establish VSOE based primarily on prior transactions
California Society of CPAs
We are concerned with the ability to estimate the extent to which that price protection will be
required ….given that much of the software industry is known for its short product cycles and
intense competition.
Inso Corporation
We believe the concept of VSOE may be difficult for software companies to apply in
practice….concept of fair value is the value a willing buyer will spend.
45
Quote from the comment letters
Respondent/Issue
Firm should defer revenues for collections beyond a 12-month period
Arthur Anderson
It is unclear….
1. revenues should be recognized on cash basis.
2. beyond 12 months be recognized in the (later) period when they become due and within 12
months recognized at delivery
3. In the period they become due within 12 months
Cash basis is inconsistent with other areas of revenue recognition (e.g. real estate ... when entire
revenue is recognized on down payment)
J. D. Edwards
Revenue would be understated if all of the vendor obligations have been met and only negotiated
payment terms remain.
Firm should defer revenues for the entire amount of coupons issued instead of accruing only the costs of expected redemption
Mysoftware Company
We would have to defer 100% of the revenues of the free products in redemption and coupon
implementations… We have been accruing the redemption costs rather than defer revenues.
Software Publishers Association
The accounting used for coupons in the retail/grocery industry should be considered whereby the
seller recognizes revenue and estimated costs of providing the additional software products are
accrued.
*I obtained copies of comment letters from the AICPA library.
46
47
APPENDIX C
DEFINITIONS OF VARIABLES
Assets
Revenue
Net income
Market value
=
=
=
=
Age (in years)
=
Implementation year
=
Prescribed implementation year
=
After
=
Return on assets
Return on equity
=
=
Revenue to assets ratio
Deferred Revenue
=
=
Normal deferred revenue ratio
=
Normal deferred revenue
=
SOP 97-2-created deferred revenue
=
DummyIncrease
=
Proforma Revenue
=
Proforma Net Income
=
Excess_BHR
=
Excess market return
=
High minus low
=
Compustat DATA6
Compustat DATA12
Compustat DATA172
Common shares outstanding (Compustat DATA25) × fiscal year
end closing price (Compustat DATA199).
Fiscal year – year of incorporation. I obtained incorporation date
from the 10-K filings.
Fiscal year in which firm starts applying the SOP 97-2 revenuerecognition rules. I obtained this information from firm’s 10-K
filings by studying the “revenue recognition” section in the
“significant accounting policies” footnote.
Firms’ fiscal years beginning after December 15, 1997. For firms
with fiscal year ending June to November, the prescribed year was
fiscal year (Compustat YEARA) 1999. For all other firms, the
prescribed year was fiscal year 1998.
Dummy variables set to 1 for observations in the prescribed
implementation year, and 0 for the year prior to that year.
Net income (Compustat DATA172) / Total assets (DATA6).
[Share price (Compustat DATA199) – Lag(Share
price)]/Lag(share price).
Revenue (Compustat DATA12) / Total Assets (DATA6).
Deferred revenue account. I obtained information on this account
from the liabilities section and the footnotes in firms’ 10-K filings.
I searched for terms including: “deferred revenue,” “unearned
income,” and “billings in excess of revenue”.
The average of Deferred Revenue/ Revenue ratio for the two years
prior to the implementation year.
Normal deferred revenue ratio × Revenue. This represents the
deferred revenue account had SOP 97-2 not been implemented.
Reported deferred revenue – normal deferred revenue. This
represents changes in deferred-revenue accounts that result from
SOP 97-2 implementation.
Dummy variables set to 1 for observations with positive SOP 972-created deferred revenues; and 0 otherwise.
Revenue + SOP 97-2-created deferred revenues. Represents
revenues had SOP 97-2 not been implemented.
Net Income + SOP 97-2-created deferred revenues. Represents net
income had SOP 97-2 not been implemented.
Excess buy and hold return [Return on equity – returns for the
risk-free securities (CRSP RF)] over the fiscal year.
[Cumulated monthly returns for value-weighted index (CRSP
VWRETD) – those for the risk-free securities (CRSP RF)] over
the fiscal year.
[Cumulated monthly returns for the high-growth portfolios – those
for the low-growth portfolios (CRSP HML)] over the fiscal year.
48
Small minus big
=
[Cumulated monthly returns for the small-cap portfolios – those
for the big-cap portfolios (CRSP SMB)] over the fiscal year.
Momentum
= [Cumulated monthly returns for the high-prior-returns portfolio
– those for the low-prior-returns portfolios (CRSP UMD)] over
the fiscal year.
7-Day_MAR
= Cumulated daily returns over the 7-trading-days period (–3 to
+3) surrounding the event days (CRSP RET) – the market
return for the same period (CRSP VWRETD).
Earnings announcement date
= Compustat RDQE for fourth quarter (Compustat QTR) of the
fiscal year.
10-K filing date
= Dates stamped on 10-K filings, obtained from SEC EDGAR
database.
All regression variables are winsorized at 5% and 95% levels, on a replacement basis.
49
TABLE 1
Sample derivation and distribution
Panel A: Sample selection
Distinct firms
H1
First Stage (Compustat)
Firms in compustat with 4-digit SIC code = 7372 or 7373 and
data available for any fiscal years 1996-2000
Less: Firms that do not have continuous data for revenue,
assets, and closing price for each of the three years ending in
the prescribed implementation year
Final sample (H1)
Less: Firms that do not have revenue and assets data for each
of the fiscal years 1996-2000
Base sample for hand collection (H2 and H3)
1,007
Distinct
firms
H2 and H3
1007
−599
408
−584
423
Second Stage (Hand-collected from10-K filings)
Less: Firms not in software industry (e.g. a holding company)
Less: Foreign filers
Less: Firms with no data on deferred revenue
Less: Firms with GVKEYs above 66,500 that have no
electronic 10-K filings in early years of the sample period
Filtered sample (H2 and H3)
−20
−37
−55
−25
286
Third stage (Implementation year can be determined)
Filtered sample
Firms for which implementation year could not be determined
Final Sample (H2 and H3)
286
−35
251
Supplementary test 2 (CRSP data is available)
Filtered sample
Firms for which CRSP data is not available
Final Sample (Supplementary test 2)
251
−39
212
TABLE 1 CONTINUED
Panel B: Distribution of sample firms
Distinct firms
Break-down of final sample
H1
Prepackaged software firms
(SIC code 7372)
302
Integrated systems design
firms (SIC code 7373)
106
Total firm-years
408
Percent of
sample
Distinct firms
H2 and H3
Percent of
sample
74%
187
75%
26%
100%
64
251
25%
100%
Panel C: Distribution by implementation fiscal-years (H2 and H3 sample)
Distinct firms
H2 and H3
Firms that adopted SOP 97-2 in the prescribed year
219
Firms that adopted early
4
Firms that adopted late
28
Total firm-years
251
Percent of
sample
87%
2%
11%
100%
Panel D: Distribution by implementation fiscal-years (corresponding to Compustat YEARA)
Implementation year
(Prescribed: for H1)
Distinct firms
Percent of
Distinct firms
(Actual: for H2 and H3)
H1
sample
H2 and H3
1997
4
1998
303
74%
181
1999
105
26%
59
2000
7
Total firm-years
408
100%
251
Percent of
sample
2%
72%
23%
3%
100%
50
51
TABLE 2
Descriptive Statistics
Panel A: Descriptive statistics for sample firms in the prescribed implementation year (H1:408
firms)
Standard
First
Third
Mean
Median
deviation
quartile
quartile
Firm demographics
Assets in $M
Revenue in $M
Market value in $M
Net income in $M
500
461
2,272
43
3,427
3,087
25,037
460
11
12
17
-5
37
36
55
0
145
139
316
6
Net income as % of lag
assets
-26%
155%
-25%
-2%
11%
27%
38%
153%
209%
-47%
-8%
-9%
14%
48%
41%
Return on equity in %
Revenue growth in %
Panel B: Descriptive statistics for sample firms in the actual implementation year (H2 and H3:251
firms)
Standard
First
Third
deviation
quartile
Median
quartile
Mean
Firm demographics
Assets in $M
Revenue in $M
Market value in $M
Net income in $M
Deferred Revenue in $M
Age in years
Net income as % of lag assets
Return on equity in %
Revenue growth in %
SOP 97-2-created deferred
revenue as % of lag assets
Percent change in deferred
revenue to revenue ratio over
the prior years’ level
337
276
1,966
32
37
12
-11%
35%
46%
1,634
1,185
17,281
301
222
9
40%
145%
195%
16
17
23
(7)
1
6
-23%
-44%
-6%
50
47
85
0
5
11
0%
-3%
18%
161
138
409
7
15
16
12%
55%
46%
1.66%
7.78%
-2.41%
0.84%
5.13%
42.1%
84.2%
-0.08%
12.2%
60.3%
TABLE 3
Change in deferred-revenue account upon SOP 97-2 implementation
Panel A: Changes in deferred-revenue account in the following four years: The two years before the implementation year, the implementation
year, and the year after the implementation year
Difference
Means
Median
t-test* Wilcoxon*
75
Percentile (p-value) (p-value)
Mean
25
Percentile
Median
Two years before the implementation year
10.0%
2.6%
7.7%
14.5%
One year before the implementation year
10.4%
3.7%
8.6%
14.5%
The implementation year
12.5%
5.1%
10.8%
17.0%
The year after the implementation year
11.8%
4.7%
9.2%
17.4%
Two years before the implementation year
10.4%
2.9%
7.9%
15.0%
One year before the implementation year
10.9%
3.7%
9.6%
15.3%
The implementation year
13.2%
6.0%
11.7%
18.4%
The year after the implementation year
13.0%
5.6%
11.5%
18.5%
Comparing individual instances
Change
Change
is
No
is
positive
change
negative
Deferred-revenue as a percent of assets
0.4%
(0.35)
2.1%
(0.01)
-0.7%
(0.45)
0.9%
(0.21)
2.2%
(0.01)
-1.6%
(0.18)
0.5%
(0.30)
2.3%
(0.01)
-0.2%
(0.85)
1.7%
(0.19)
2.1%
(0.01)
-0.2%
(0.39)
47%
0%
53%
66%
0%
34%
48%
0%
52%
46%
0%
54%
67%
0%
33%
49%
0%
51%
Deferred-revenue as a percent of revenue
* One-tailed tests
52
TABLE 3 CONTINUED
Panel A continued
Difference
Means
Median
t-test* Wilcoxon*
75
Percentile (p-value) (p-value)
Mean
25
Percentile
Median
Two years before the implementation year
5.2%
0.9%
2.7%
6.6%
One year before the implementation year
6.0%
1.4%
3.3%
8.8%
The implementation year
9.4%
1.8%
5.5%
11.6%
The year after the implementation year
6.9%
1.1%
3.0%
8.8%
Comparing individual instances
Change
Change
is
No
is
positive
change
negative
Deferred-revenue as a percent of market value
0.8%
(0.12)
3.4%
(0.01)
-2.9%
(0.01)
1.6%
(0.06)
2.2%
(0.01)
-2.5%
(0.01)
53%
0%
47%
65%
1%
34%
33%
2%
65%
* One-tailed tests
53
TABLE 3 CONTINUED
Panel B: Changes in deferred revenue by industry sub-classification
Mean
25
Percentile
Median
Difference
Means
Median
t-test* Wilcoxon*
75
Percentile (p-value) (p-value)
Comparing individual instances
Change
Change
is
No
is
positive
change
negative
Deferred-revenue as percent of revenue for prepackaged software firms (SIC code 7372), which assume less complex multiple-element obligations
Two years before the implementation year
One year before the implementation year
The implementation year
11.5%
4.3%
10.3%
15.8%
12.1%
4.9%
10.8%
17.6%
14.2%
7.3%
13.0%
18.6%
0.6%
(0.36)
2.1%
(0.01)
-0.1%
(0.95)
0.5%
(0.23)
2.2%
(0.01)
-0.2%
(0.94)
48%
0%
52%
65%
0%
35%
49%
0%
14.1%
7.2%
12.8%
19.1%
Change in deferred revenue to revenue ratio (implementation year) as a percent of previous year’s ratio = 36.3% significant at p-level of 0.01
The year after the implementation year
51%
Deferred-revenue as percent of revenue for integrated systems design firms(SIC code 7373), which assume more complex multiple-element obligations
Two years before the implementation year
One year before the implementation year
The implementation year
7.1%
1.1%
4.2%
7.4%
7.0%
1.5%
4.1%
9.9%
10.0%
3.1%
6.4%
13.5%
-0.1%
(0.95)
3.0%
(0.02)
-0.3%
(0.85)
-0.1%
(0.69)
2.3%
(0.03)
-0.5%
(0.63)
40%
0%
60%
73%
0%
26%
52%
0%
9.7%
2.3%
5.9%
14.4%
Change in deferred revenue to revenue ratio (implementation year) as a percent of previous year’s ratio = 60.5% significant at p-level of 0.01
The year after the implementation year
48%
Differential effect of SOP 97-2 on the two industry subgroups
Change in deferred revenue to revenue ratio (implementation year) as a percent of previous year’s ratio = 24.2% significant at p-level of 0.03
* One-tailed tests
54
55
TABLE 4
Whether earnings informativeness declines upon SOP 97-2 implementation
Panel A: Earnings Response Coefficient (ERC) compared for the following two years: the
year before the prescribed implementation year and the prescribed implementation year
Excess_BHRit = β01 + β01 × After
+ β11 × ∆NetIncomeit + β12 × After × ∆NetIncomeit
+ β21 × ∆Revenueit
+ β22 × After × ∆Revenueit
+ Σ βs1 × Controlsit
+ Σ βs2 × After × Controlsit
Variable
Predicted sign
Intercept
After (β02)
Change in net income
+
−
After × Change in net income (β12)
Change in revenue
+
−
After × Change in revenue (β22)
Controls for Fama-French factors, Momentum, Size, and
Year- and Sub-industry fixed effects
Total number of observations (N)
Number of distinct firms
F-Value
Probability
Ftest*: β02 + β12 = 0
Ftest*: β02 + β22 = 0
All dollar denominated variables are deflated by beginning assets.
* one-tailed tests for directional hypotheses
+ εit
Estimate
0.638
-1.019
0.446
-0.241
0.504
0.025
p-value*
0.222
0.219
0.001
0.094
0.001
0.856
Yes
816
408
8.35
0.001
0.067
0.122
TABLE 5
Whether the market considers the SOP 97-2-created deferred revenues as revenues or as current liability
Panel A: The implementation year
Excess_BHRit = β0 + β1 × ∆DeferredRevenueit + β2 × ∆Revenueit + β4 × ∆NetIncomeit + Σ βs × Controlsit + εit
(1)
Excess_BHRit = β0 + β1 × ∆NormalDeferredRevenueit + β2 × SOP97-2CreatedDeferredRevenueit + β3 × ∆Revenueit + β4 × ∆NetIncomeit
+ Σ βs × Controlsit + εit
(2)
Excess_BHRit = β0 + β1 × ∆NormalDeferredRevenueit + β2 × ∆ProformaRevenueit + β4 × ∆ProformaNetIncomeit + Σ βs × Controlsit + εit
(3)
(1)
(2)
(3)
Reported accounting
Disaggregated deferredProforma accounting
numbers
revenue account
numbers
Predicted
Variable
sign
Estimate
p-value*
Estimate
p-value*
Estimate
p-value*
Intercept
-0.368
0.564
-0.372
0.555
-0.391
0.532
+
Change in deferred revenue
0.155
0.378
+/−
-0.711
0.153
Change in normal deferred revenue (β1)
-0.776
0.155
+
SOP 97-2-created deferred revenue (β2)
1.138
0.074
+
0.750
0.001
Change in revenue
(β3)
0.774
0.001
0.622
0.001
0.617
0.001
Change in net income
+
0.754
0.001
Proforma-change in revenue
+
Proforma-change in net income
+
0.606
0.001
Controls for Fama-French factors, Momentum, Size, Age, and
Year- and Sub-industry fixed effects
Yes
Yes
Yes
Total number of observations (N)
251
251
251
21.0%
Adjusted R-squared
17.4%
21.3%
F-Value
4.92
5.58
6.03
Probability
0.001
0.001
0.001
Second Specification:
Ftest*: β1 = β2
0.006
Ftest*: β2 = β3
0.333
First and Third Specifications:
Vuong-test*: R-square using reported numbers (1) = R-square using proforma numbers (3)
0.072
All dollar denominated variables are deflated by beginning assets.
* One-tailed tests for directional hypotheses
56
TABLE 5 CONTINUED
Panel B: The year after the implementation year
Excess_BHRit+1 = β0 + β1 × ∆DeferredRevenueit + β2 × ∆Revenueit+1 + β3 × ∆NetIncomeit+1 + Σ βs × Controlsit+1 + εit+1
(1)
Excess_BHRit+1 = β0 + β1 × ∆NormalDeferredRevenueit + β2 × SOP97-2CreatedDeferredRevenueit + Σ βs × Controlsit+1 + εit+1
(2)
Excess_BHRit+1 = β0 + β1 × ∆NormalDeferredRevenueit + β2 × SOP97-2CreatedDeferredRevenueit + β3 × ∆Revenueit+1 + β4 × ∆NetIncomeit+1
+ Σ βs × Controlsit+1 + εit+1
(3)
(1)
(2)
(3)
Disaggregated
implementation year’s
deferred-revenue
Aggregated
Disaggregated
account, controlling for
implementation year’s
implementation year’s
subsequent year’s
deferred-revenue
deferred-revenue account
earnings
account
Predict
Variable
ed sign
Estimate
p-value*
Estimate
p-value*
Estimate
p-value*
Intercept
-0.693
0.367
0.071
0.933
-0.480
0.536
+/−
Change in deferred revenue(t)
-0.514
0.470
Change in normal deferred revenue(t)
+/−
-0.488
0.573
-0.471
0.555
−
SOP 97-2-created deferred revenue(t)
-0.640
0.597
-1.939
0.043
Change in revenue(t+1)
+
0.292
0.033
0.305
0.028
Change in net income(t+1)
+
1.538
0.001
1.599
0.001
Controls for Fama-French factors, Momentum, Size, Age, and
Year- and Sub-industry fixed effects
Yes
Yes
Yes
Total number of observations (N)
251
251
251
9.81
5.82
9.26
F-Value
Probability
0.001
0.001
0.001
All dollar denominated variables are deflated by beginning assets.
* one-tailed tests for directional hypotheses
57
TABLE 6
Whether the stock prices decline when firms report increases in deferred-revenue accounts upon SOP 97-2 implementation
7-dayReturnit = β01 + β02× DummyIncreaset + β1 × ∆NormalDeferredRevenueit
+ β21 × SOP97-2CreatedDeferredRevenueit + β22 × DummyIncreaset × SOP97-2CreatedDeferredRevenueit
+ β3 × ∆Revenueit + β4 × ∆NetIncomeit + β5 × 7-dayMarketReturnt + Σ βs× Controlsit + εit
Predicte
d sign
Variable
Intercept
+/−
−
Dummy Increase (β02)
Change in normal deferred revenue
+/−
Rules-created deferred revenue
+/−
Dummy Increase × SOP97-2-created deferred revenue (β22)
−
Change in revenue
+
Change in net income
+
7-dayMarketReturn
Controls for Size, Age, and Year- and Sub-industry fixed effects
Total number of observations (N)
Adjusted R-squared
F-Value
Probability
All firm-specific variables are deflated by beginning assets.
7-trading days
surrounding 10-K filing
dates
Estimate
p-value
-0.183
0.047
0.026
0.337
-0.010
0.935
-0.512
0.136
0.686
0.103
0.055
0.041
0.034
0.186
1.704
0.001
Yes
212
9.8%
2.760
0.001
7-trading days
surrounding earnings
announcement dates
Estimate
p-value
0.123
0.144
0.021
0.418
0.004
0.970
0.205
0.512
-0.190
0.624
0.001
0.485
-0.038
0.137
0.613
0.030
Yes
212
0. 9%
1.150
0.322
58
TABLE 7
Effects of SOP 97-2 – representative disclosures from firms’ 10-K filings, before and after SOP 97-2 implementation
Panel A: Firms’ discussions on the expected effects of SOP 97-2 before implementation (SAB 74 compliance)
Compustat
Name of firm
fiscal year
Quote from the 10K filings
Most firms expected no significant effect (typical response, almost 90% of firms)
Pegasystems Inc.
1997
Beginning in 1998, the Company will be required to adopt the provisions of SOP 97-2, "Software
Revenue Recognition." The implementation of the statement is not estimated to have a significant
impact on the Company.
Some firms expressed uncertainty on new rules or their effects
Citrix Systems Inc.
1997
……. the impact on the future financial results of the Company is not currently determinable.
HNC Software Inc.
1997
…. could lead to unanticipated changes in the Company's current revenue recognition practices, and
such changes could be material to the Company's financial statements.
Some firms expected significant deferral of revenues
Ascential Software Corp
1997
The implementation may, in certain circumstances, result in the deferral of software license
revenues that would have been recognized upon delivery of the related software under preceding
accounting standards.
Veritas Software Corp.
1997
The criteria for recognizing revenue under SOP 97-2 are generally more rigorous than the previous
accounting standard and, in some cases, significantly more rigorous.
Acme Comm Corp
1997
This standard is expected to result in more conservative revenue recognition on software
transactions than was allowed under previous guidance.
A few firms planned to change their contracts to align with SOP 97-2 rules
Global Med
1997
Management's current plans to address the revenue recognition requirements of SOP 97-2 include
Technologies Inc.
substantial revisions to Wyndgate's current standard terms and conditions included as part of
Wyndgate's SAFETRACE(R) software license agreements.
A few firms expected that implementing SOP 97-2 rules would adversely affect firm’s fund raising plans
Ascential Software Corp.
1997
Failure to effectively address the revenue recognition requirements of SOP 97-2 and future revenue
recognition guidance regarding the software industry will have a material adverse affect on the
Company's revenues, gross margins and operating results. As a result, future capital raising efforts
may also be adversely affected.
59
TABLE 7 CONTINUED
Panel B: Firm’s discussion on the effects of SOP 97-2 after implementation (MD&A rules compliance)
Compustat
Name of firm
fiscal year
Quote from the 10K filings
Most firms found no material effect (typical response, almost 90% of firms)
Nyfix Inc.
1998
The Company adopted SOP 97-2 in 1998 and the effect was not material to the Company’s
operations or financial position taken as a whole.
Some firms mentioned that implementing SOP 97-2 rules led to deferral of revenues
Electronic Arts Inc.
1998
The implementation has, in certain circumstances, resulted in the deferral of certain revenues
associated with the Company's revenue promotions and products with multiple deliverable
elements
Intrusion Inc.
1998
The most significant impact of SOP 97-2 on the Company's revenue recognition accounting policies
is that for software contracts with multiple elements, revenue will generally be recognized later than
under past practices under SOP 91-1.
Intellisync Corp.
1999
The implementation has, in certain circumstances, resulted in the deferral of software license
revenues that would have been recognized upon delivery of the related software under prior
accounting standards.
A few firms blamed SOP 97-2 for their poor performance
Aspyra Inc.
1998
Although the Company has been profitable in each of its last six previous fiscal years, its 1998 fiscal
year resulted in a loss primarily attributable to the implementation of a new accounting method,
SOP 97-2
Softech Inc.
1998
Due to stricter requirements for recognizing revenue from the sale of software products,
implementation of the provisions of SOP 97-2 had the effect of reducing revenue in fiscal 1999;
however, quantifying the impact of this change in accounting was not practical.
60
TABLE 7 CONTINUED
Panel B continued
Name of firm
Compustat
fiscal year
Quote from the 10K filings
Very few firms quantified the SOP 97-2 implementation effects
Reynolds & Reynolds
1998
The implementation of this pronouncement reduced the Automotive Group's computer systems
products revenues $17,936, gross profit $11,205, operating income $10,624 and net income $6,204
or $.08 per diluted share during the six months ended March 31, 1999
ASA International Ltd.
1998
The effect of adopting SOP 97-2 on the 1998 Statement of Operations was to decrease income
before income taxes and net income by approximately $446,000 and $174,000, respectively.
Clickaction Inc.
1998
Under SOP 97-2, the Company is required to defer revenue related to certain promotional product
offerings based on the relative fair value of undelivered products included in the promotional
offering. As of December 31, 1998, the Company recorded approximately $599,000 of deferred
revenue related to such promotional product offerings
A few firms restated their accounts due to late implementation of SOP 97-2 rules
Cape Systems Group Inc.
1998 and 1999
The Company has adjusted its accounting for revenue recognition of certain software license fees
and related post contract customer support revenue for the years ended July 31, 1999 and 1998
because the Company determined that information with respect to separate pricing for each of the
multiple elements is not available.
Peritus Software
1998
The balance of the 1998 restatements relate primarily to the Company's incorrect interpretation of
Services Inc.
the complex provisions regarding recognition of revenue for combined software/services
arrangements under Statement of Position ("SOP") 97- 2, "Software Revenue Recognition,"
61
62
VITA
Name:
Anup Srivastava
Address:
Department of Accounting
Mays Business School
460 Wehner Building
Texas A&M University
College Station, Texas 77843-4353
Email Address:
anupsrivastava@yahoo.com
Education:
B.Tech, IIT Delhi, 1988
M.B.A., University of Delhi, 1990
Ph.D., Texas A&M University, 2008
Publications:
1. McAnally, M. L., Srivastava, A., and Weaver, C. 2008. “Executive stock options,
missed earnings targets and earnings management.” The Accounting Review 83(1):
185-216.
2. Efendi, J., Srivastava, A., and E. P. Swanson. 2007. “Why do corporate managers
misstate financial statements? The role of in-the-money options and other
incentives.” The Journal of Financial Economics 85(3): 667-708.
Scholarships:
•
•
•
•
Deloitte Fellowship of $ 25,000
Regents’ Scholarship
Mays Business School Dean’s Scholarship for meritorious performance
National Talent Search Scholarship awarded by Government of India